Introduction

Commercial white light emitting diodes (LEDs) are predominantly based on InGaN blue light emitting diodes with an emission peak at 450 nm. The emitted blue light passes an active layer containing a phosphor, which converts a part of the blue light of the LED to yellow light. The combination of the transmitted blue light and converted yellow light appears white to the human eye. Up to now, mostly Ce³⁺:YAG (Ce³⁺ doped yttrium aluminum garnet Al₅Y₃O₁₂) is used as yellow phosphor. Alternative phosphors as e. g. Ce³⁺:LuAG (Ce³⁺ doped lutetium aluminum garnet Al₅Lu₃O₁₂) provide different emission spectra and can therefore help to extend the emitted spectrum of the white LED and in combination with red light emitting phosphors can improve their color rendering index. The phosphor is commonly embedded into a polymer, usually a polysiloxane which is stable at temperatures up to around 200 °C. This phosphor-polymer composite shows a strong absorption in the blue range and a broad re-emission between 500 and 700 nm, centered at around 550-560 nm if Ce³⁺:YAG is used. Ce³⁺:LuAG emits at slightly lower wavelengths with a peak at around 510 nm. The polysiloxane polymers show a very limited thermal conductivity [1] and degrade due to heat and irradiation with high power short wavelength light [2, 3]. Hence, especially for high power LEDs, the heat generated in the Ce³⁺:YAG phosphor cannot effectively be removed promoting further degradation of the fluorescent layer and the LED chip. This strongly limits the maximum light intensity of the LEDs. Other encapsulation materials as e. g. epoxy resin are even more sensitive to heat and short wavelength irradiation [2, 4, 5].

For high power applications, two possibilities to overcome these restrictions have been proposed: firstly the preparation of a glass doped with yttrium, aluminum and cerium and subsequent crystallization of Ce³⁺ doped YAG directly from the glass. This route has been realized, but requires high expenditures and usually temperatures in the range from 800 to 1100 °C for the crystallization process which usually are too high to guarantee dimensional stability during thermal treatment [6–10]. For example, Tarafder et al. recently reported on the preparation and properties of RE-doped YAG glass-ceramics in the K₂O/Y₂O₃/Al₂O₃/SiO₂ system prepared by a controlled crystallization technique [10]. Furthermore, Keshavarzi et al. reported on the crystallization of YAG from silicate glasses with small alkali concentrations and without any alkali [7, 8]. Preparation of Ce³⁺:YAG glass ceramics in the system Y₂O₃/Al₂O₃/SiO₂ is also reported by Fujita et al. [6]. Usually, silicate glasses which contain the two required components, Y₂O₃ and Al₂O₃ do not show a high tendency towards crystallization and require fairly high crystallization temperatures or, however, precipitate other crystalline phases, such as yttrium silicates and not YAG.

The other possible route is to mix the phosphor with an industrially produced glass powder, pressing the mixture and subsequent sintering [11]. Both routes enable to embed the YAG crystals directly into a glass matrix, which provides a higher thermal conductivity, higher stability against short wavelength irradiation and protection against any kind of corrosion. Concerning the second route, several approaches have been proposed during the past few years. Mostly glasses with high refractive indices and low T_g are used; on the one hand to match the refractive index to the phosphor to limit light scattering and, on the other hand, to allow low sintering temperatures in order to minimize chemical reactions of phosphor and glass. These glass compositions include tellurite [12, 13, 16, 17], antimony borate [13, 14], lead boro-silicate [15], barium boro-silicate [18] and zinc boro-silicate glasses [19, 20]. Tin phosphate glasses with extremely low T_g might also be interesting embedding materials, especially for thermally unstable phosphors [21]. In some cases highly transparent glass ceramic samples could be obtained using this method [14, 16]. However, transparent glass ceramics might not be necessary for white LED lighting applications. The polymer embedded Ce³⁺:YAG layers are not transparent either. Furthermore, most of the above mentioned glass compositions are not suitable for mass production. A different approach is the use of low cost, readily available glasses as e. g. window / container glass (soda-lime silica glass) for embedding of the Ce³⁺:YAG phosphor. Some successful experiments of Ce³⁺:YAG embedding into similar glass compositions in the system Na₂O/CaO/Al₂O₃/SiO₂ are reported in [1, 22, 23]. This paper presents a route for the preparation of low cost Ce³⁺:YAG and Ce³⁺:LuAG glass ceramics by using soda-lime container glass as well as an analysis of the chemical stability of the phosphor under different sintering conditions and efficiency measurements of the produced glass ceramics.

Experimental procedure

Commercially produced Ce³⁺ doped yttrium aluminum and lutetium aluminum garnet powders (NEMOTO & CO., Ltd., Japan) with grain sizes of 15-18 and 12 µm respectively, were used. As glass, a soda-lime silicate (NCS) glass with the molar composition 15.5 Na₂O/10.7 CaO/73.8 SiO₂ was used. The glass was prepared by melting a batch of 250 g at 1450°C. After melting for 2 hours the melt was cast into cold water to obtain coarse grained precursor glass powder. Subsequently, the powder was dried and later powdered to grain sizes < 30 µm. Then the glass powder and the phosphor were mixed, homogenized and the mixture given to platinum-gold molds and then heated to temperatures in the range from 800 to 1000 °C, using a rate of 120 K/min and subsequently kept at this temperature for 10 to 30 min. The Ce³⁺:YAG/NCS specimens had an intense yellow color and the Ce³⁺:LuAG/NCS samples an intense yellow greenish color (Fig. 1). There was no difference in the color of samples produced with different sintering parameters.

 

Fig. 1 Ce³⁺:YAG/NSC (left) and Ce³⁺:LuAG/NCS (right) composites prepared at 1000 °C / 10 min., the diameter of the samples is 12 mm.

Download Full Size | PDF

The sintered specimens were ground to a thickness of 450 µm and subsequently polished. The silicone embedded samples were produced using a teflon coated mold. The fluorescence excitation and emission spectra were recorded using a Shimadzu RF-5301 PC spectrometer in the wavelength range of 220 to 900 nm (spectral resolution: 0.2 nm). These measurements were conducted in transmission mode, illuminating the samples at the back and measuring the fluorescence emission at the opposite side. The samples were further characterized by X-ray diffraction (Siemens D5000) using Cu K_α radiation. The microstructures of the specimen were studied with a field-emission scanning electron microscope Jeol JSM 7001F. The efficiency measurements have been conducted using an integrating sphere and an array spectrometer CAS 140CT (Instrument Systems GmbH, Germany).

Results and discussion

Figure 2 shows the XRD-patterns of two sintered specimens (1000 °C / 10 min) containing the respective phosphors. The XRD pattern of the YAG containing sample is in perfect agreement with the JCPDS file nr. 01-073-1370 (Al₅Y₃O₁₂) and does not show any additional lines which give a hint at the occurrence of an additional crystalline phase. The pattern of the LuAG containing sample is very similar to the YAG pattern but slightly shifted due to the smaller lattice constants and ionic radii of Lu³⁺ (86.1 pm) in comparison to Y³⁺ (90 pm). The lines are all in agreement with the JCPDS file nr. 01-073-1368 (Al₅Lu₃O₁₂).

 

Fig. 2 XRD-patterns of the sintered Ce³⁺:YAG/NSC and Ce³⁺:LuAG/NCS composites prepared at 1000 °C / 10 min.

Download Full Size | PDF

Figure 3 shows the fluorescence excitation (left) and emission spectra (right) of Ce³⁺:YAG/NCS samples in two different spectral ranges. The emissions observed at around 380 nm are attributed to Ce³⁺ in the glassy matrix [24], while the emissions at around 550 nm are due to Ce³⁺ incorporated into the YAG crystal lattice. Notice that different excitation wavelengths have been used; 320 and 476 nm respectively (see also left part of Fig. 3). The difference in the spectra can be used to evaluate the chemical stability of the Ce³⁺:YAG powder under different sintering conditions: While the fluorescence emission of the Ce³⁺ doped glass is very small for a sintering temperature of 800 °C, it increases strongly if the sample is sintered at 1000 °C for 10 min. An even higher fluorescence emission of the Ce³⁺ ions in the glass matrix is observed in the sample sintered at 1000 °C for 30 min. The fluorescence intensity of Ce³⁺ incorporated in the YAG matrix is maximal in the sample sintered at 1000 °C for 10 min and decreases by 15% for the sample sintered at 1000 °C for 30 min. The fluorescence of the sample sintered at 800 °C for 10 min is close to the 1000 °C / 10 min sample. These spectra clearly show that the Ce³⁺:YAG particles are dissolved by the NCS-glass during the sintering process. As expected, a higher quantity of phosphor is dissolved with increasing sintering temperature and sintering time. Chen et al. report very similar effects for Ce³⁺:YAG/glass samples prepared at 700 to 900 °C using high-resolution transmission electron microscopy [23]. For the sample prepared at 900 °C, a diffusion zone of about 40 nm between the Ce³⁺:YAG particles and the glass matrix was found, while there was no such diffusion zone for the sample sintered at 700 °C. However, the different emission spectra of the Ce³⁺:glassy phase were not measured in this case. Here, the ratio of the peak intensities of both Ce³⁺ populations could provide an easy-to-obtain measure of the stability of the phosphor in different glass matrixes and under different sintering conditions. Especially interesting is the effect of the chemical composition of the glass. In an additional experiment Ce³⁺:YAG powder was embedded into a borosilicate glass with similar T_g at 1000 °C for 10 min. For this sample the emission at 550 nm (Ce³⁺:YAG phase) was notably lower, and the intensity of the Ce³⁺:glass phase was about 4 times stronger than for the NCS-glass embedded samples resulting in an intensity ratio of both phases of almost 1:1 (not shown). Zhou et al. investigated the effect of different sintering temperatures using tellurite glass for embedding of Ce³⁺:YAG. After sintering at a temperature of only 700°C for 30 min the samples turned almost colorless indicating a nearly complete dissolution of the Ce³⁺:YAG phosphor by the tellurite glass [17]. However, the spectra of the Ce³⁺ ions in the glass phase have not been measured. Anyway, these results clearly show the effect of the chemical composition of the embedding glass on the stability of the embedded phosphor. Therefore low-T_g glasses are not necessarily a good choice as embedding material.

 

Fig. 3 Fluorescence excitation (left) and emission spectra (right) of Ce³⁺:YAG/NCS composites produced under different conditions at excitation/emission wavelengths of 320/382 and 476/543 nm respectively.

Download Full Size | PDF

Figure 4 shows the excitation and emission spectra of the Ce³⁺:LuAG and Ce³⁺:glass phases for Ce³⁺:LuAG/NCS specimens in analogy to Fig. 3. While the observed Ce³⁺:LuAG spectra are different in their shapes and spectral position compared to those of the Ce³⁺:YAG containing specimens, the spectra of the Ce³⁺ in the glass phase show almost the same shape and position as the spectra in Fig. 3. However, the Ce³⁺:LuAG phosphor seems to be slightly more stable than the Ce³⁺:YAG. This is indicated by the notably lower emission of the Ce³⁺:glass phase of the 1000 °C / 10 min sample and the higher emission at 510 nm for the 1000 °C / 30 min sample (Figs. 3 & 4 are equally scaled).

 

Fig. 4 Fluorescence excitation (left) and emission spectra (right) of Ce³⁺:LuAG/NCS composites produced under different conditions at excitation/emission wavelengths of 320/381 and 454/507 nm respectively.

Download Full Size | PDF

The samples have also been studied by electron microscopy. As an example, Fig. 5 shows an SEM micrograph of the Ce³⁺:YAG/NCS sample prepared at 1000 °C / 30 min. For samples of different preparation temperatures / times, no significant differences are observed. Interestingly the samples contain only few pores.

 

Fig. 5 SEM micrograph of the Ce³⁺:YAG/NCS composite prepared at 1000 °C / 30 min.

Download Full Size | PDF

In order to study the effect of the phosphor concentration on the fluorescence properties, samples with different Ce³⁺:YAG concentrations were sintered. For that purpose, specimens with concentrations of 10, 17.5, 25 and 31 wt% Ce³⁺:YAG were prepared and measured with a fluorescence spectrometer in transmission mode. As in the measurements before, an excitation wavelength of 476 nm had been used. The attributed fluorescence emission spectra are shown in Fig. 6. All samples exhibited the same thickness of 450 µm. Surprisingly, the concentration of the phosphor had only a minor effect on the fluorescence intensity and the intensities agree with each other within a limit of only 4%. But it should be mentioned that the chromaticity coordinates of these samples are, however, different.

 

Fig. 6 Fluorescence emission spectra of samples with different wt% concentrations of Ce³⁺:YAG phosphor (excitation wavelength 476 nm).

Download Full Size | PDF

An explanation for this surprising effect could be that although a higher percentage of blue light is converted to yellow light by samples of higher Ce³⁺:YAG concentrations also more light is scattered by these samples, counterbalancing the increased fluorescence emission intensity and resulting in a mostly constant fluorescence emission intensity under the measuring conditions which were used here. Due to the measurement in a spectrometer only photons emitted in direction of the detector are counted; most of the scattered photons are ignored under these conditions. Additional experiments with samples of constant Ce³⁺:YAG concentration (10 wt%) but different sample thickness support this assumption: For relatively thin samples the fluorescence intensity rises with increasing thickness of the samples. For samples of higher thickness the intensity increase is getting smaller and smaller due to increased scattering. Even at low phosphor concentrations this effect is already obvious at a samples thickness of less than 1 mm. Further experiments with varying excitation intensities showed no different result (except an altered emission intensity of all samples). Therefore an influence of the intensity of the excitation light, e. g. a generally too low excitation intensity in the spectrometer, could be ruled out.

Figure 7 shows the efficiencies of Ce³⁺:YAG and Ce³⁺:LuAG samples in comparison to samples composed of the respective phosphor embedded into a silicone matrix. These measurements have been conducted using an integrating sphere and samples sintered at 1000 °C / 10 min. The efficiencies in lm/W are plotted against the CIE x-value. The radiant power emitted from the blue LED (surface emitting type) was measured beforehand with a near hemispherical encapsulation which was not cured. This setup was used to normalize the light flux output to 1 W of blue LED light input to the phosphor plates. After removing the transparent encapsulation the phosphor probes where placed on the LED chip, again not cured to have equal refractive index, and the light flux of the phosphor converted LED was measured. Using this setup the efficiency of the conversion layer can be measured independently of the LED chip and the unit lm/W refers to lumen per Watt blue radiant power emitted by the blue LED. Therefore our values are higher in comparison to other publications which use lumen per Watt of electrical power supplied to the LED chip. The CIE x-value was varied by altering the sample thickness. It is seen that the efficiencies increase with increasing x-value, which is related to a higher conversion rate of blue light to better perceptible yellow light. This is observed for the Ce³⁺:YAG/NCS and the Ce³⁺:LuAG/NCS specimens and also for the samples composed of phosphors embedded into the silicone matrix. The efficiencies of the Ce³⁺:LuAG samples are approximately at the same line as the polymer matrix samples. That means, the quantum efficiency of the prepared phosphor/glass samples is almost as high as for the polymer matrix samples. In the case of the Ce³⁺:YAG/glass samples slightly smaller values were measured for the phosphor/glass samples than for their polymer embedded counterparts. These results strengthen the assumption that the Ce³⁺:LuAG phosphor is chemically more stable than the Ce³⁺:YAG.

 

Fig. 7 Efficiency of Ce³⁺:YAG (left) and Ce³⁺:LuAG (right) containing samples as a function of their CIE x-value for glass sintered specimens (squares) and for silicone matrix embedded phosphors (circles).

Download Full Size | PDF

Conclusion

Commercial Ce³⁺:YAG and Ce³⁺:LuAG phosphors were embedded into a soda-lime silicate glass with the molar composition 15.5 Na₂O / 10.7 CaO / 73.8 SiO₂. The samples were sintered at temperatures in the range from 800 to 1000 °C for 10 and 30 min. Fluorescence spectra clearly show a dissolution of the phosphors during the sintering process, especially at 1000 °C. However, if the processing time is low (10 min) almost no fluorescence intensity is lost in comparison to samples sintered at 800 °C. The emission intensity of Ce³⁺ ions in the glassy phase can easily be used as a measure of this dissolution process, since the Ce³⁺:NCS-glass emission peak is in a much different spectral range than the emission of the Ce³⁺:YAG and Ce³⁺:LuAG phosphors.

Different glass compositions can be chemically much more aggressive and dissolve the phosphor more easily under the same sintering conditions, but also at even lower sintering temperatures and shorter times. Therefore low-T_g glasses are not necessarily a good choice as embedding material.

The phosphor concentration was varied between 10 and 31 wt% and had almost no effect on the fluorescence intensity of 450 µm thick samples. Most likely the increased fluorescence emission due to higher phosphor concentrations is counterbalanced by more intense light scattering in the samples.

Efficiency measurements show slightly lower values for the phosphor/glass samples than their polymer embedded counterparts. In general the Ce³⁺:LuAG based samples showed a higher efficiency indicating a higher chemical stability of this phosphor.